TI has a good filter designer too, which lets you play with the Q factor a bit (i.e. choose filters that don't drop off so quick) - https://webench.ti.com/filter-design-tool/filter-response
Nice preamp but yours has ~100k input impedance! ... not what is being measured here!
Sometimes I also get exasperated and go to the "trivial solution" 😉 , kicking the chess table and spreading pawns and kings all over the place 🙂
Pity this is a very nice looking board, chock full of parts, big main caps, relay, Servo biasing (I guess) , "good" Op Amps, the works.
If we had a schematic (ask the builder, who knows? You might get lucky) we could relatively easily do a lobotomy, meaning remove/bypass the complex front end and go straight to LM3886 input pin 🙂
In principle keeping all the other goodies, such as nice power supply, heatsinking, presumed speaker DC protection, etc.
Without schematic it becomes impossible.
I mean: 4 (FOUR) Op Amps per LM3886? 😱
I didn't mean the values were the same, just that the layout is likely similar, specifically with the biasing that was going on with the two identical resistors. But yah I agree!
So, schematic. I sat down again (I do that a lot lately) and put together the preamplification stage in what is hopefully human-readable form. KiCad is awesome. 👍
This is the stock configuration (please disregard the missing NE5334). Does this look like something that will work as a preamplifier with let's say something like 4.5x gain and a low input impedance of around 2.5KΩ; Or am I just pasting what the cat dragged in? In my initial "schematic" efforts I had missed a crucial connection to the inverting input of the op amp, and also labeled a resistor as 2.8KΩ instead of 8.2KΩ which was the actual value. The op amp power supply is obviously ommited. I'm trying to see if I got everything right or still missing connections or other elements of the circuit.
Btw, @JMFahey , the LM3886 pin is connected to the right of the bottom capacitor, it just takes a wormhole on the board to get there. I didn't expect to see the LM3886 right after this stage as there's circuitry that seemed like an intermediate stage between preamp and poweramp, but obviously it serves other purposes.
Any comments, suggestions, corrections are more than welcome. If I get this straight I will follow up with the aforementioned magic jumper kind of sorcery.
Schematic update. Series R1+R3 didn't make much sense, so I discover their connection is grounded. I may be missing more ground points.
Don't know if this Q has been resolved, but it is very easy.
Drive the input directly from a signal generator and measure Vout.
Then put a series R between the signal generator and input and change that resistor value until you have half the output from the 1st step.
Now Rin is the value of the series resistor.
Jan
Drive the input directly from a signal generator and measure Vout.
Then put a series R between the signal generator and input and change that resistor value until you have half the output from the 1st step.
Now Rin is the value of the series resistor.
Jan
Thanks for your input Jan. I have used this method to estimate Zin, which I would like to somehow increase.
Since you mentioned a signal generator, let's say a sine signal, the question arises: at what frequency do you measure? Next amplifier specification I will come across will mention Zin, for example 47KΩ, but no frequency where that is measured. Is there a standard frequency (I suspect 1KHz) that is implied when mentioning impedance in the world of audio?
Last edited:
Yes, normally you measure at 1kHz for convenience. Although Zin does sometimes vary with f, it should not be too much.
After all, you're not interested in Zin down to 0.1%, you just need to know if it's sufficient to be driven by the source conveniently.
I don't know of any preamp or DAC and such that would have an issue driving nominally 47k. Is that what you measured?
And why would you like to increase it?
But nothing stops you from checking at 20Hz and 20kHz.
Jan
After all, you're not interested in Zin down to 0.1%, you just need to know if it's sufficient to be driven by the source conveniently.
I don't know of any preamp or DAC and such that would have an issue driving nominally 47k. Is that what you measured?
And why would you like to increase it?
But nothing stops you from checking at 20Hz and 20kHz.
Jan
He measured around 2k ohm Z in, which is surprisingly low.
He also found amp is extremely sensitive , requiring very low volume settings.
I suspected a "mic level" input but product page says nothing about that, yet measurements ....
We will have to wait for a schematic, there is 4 Op Amps per LM3886 there, it´s not a straight connection by any means.
Personally, if I had it on my bench, I would search for LM3886 pin 10 (+ IN), disconnect anything there, add a 47k resistor to ground (using same pad where NFB returns) so it as a clean ground reference, and add a short wire (I guess less than 10 cm) to input connector, bypassing anything in between.
Obviously also removing input connector connections to internal preamp too.
Bet many other experienced LM3886 builders around here can do the same, they are used to identifying pins and following tracks.
The total job could be as simple as removing 2 or 3 components, adding the 47k resistor and a piece of wire, so restoring normal sensitivity, high-ish impedance, etc. and keeping all other goodies present in that nicely made PCB.
But I hesitate to suggest that to a relatively inexperienced user, specially having NO schematic at all.
He also found amp is extremely sensitive , requiring very low volume settings.
I suspected a "mic level" input but product page says nothing about that, yet measurements ....
We will have to wait for a schematic, there is 4 Op Amps per LM3886 there, it´s not a straight connection by any means.
Personally, if I had it on my bench, I would search for LM3886 pin 10 (+ IN), disconnect anything there, add a 47k resistor to ground (using same pad where NFB returns) so it as a clean ground reference, and add a short wire (I guess less than 10 cm) to input connector, bypassing anything in between.
Obviously also removing input connector connections to internal preamp too.
Bet many other experienced LM3886 builders around here can do the same, they are used to identifying pins and following tracks.
The total job could be as simple as removing 2 or 3 components, adding the 47k resistor and a piece of wire, so restoring normal sensitivity, high-ish impedance, etc. and keeping all other goodies present in that nicely made PCB.
But I hesitate to suggest that to a relatively inexperienced user, specially having NO schematic at all.
A schematic would be really useful, but I don't think we need it now (and in any case I think the seller either doesn't have it or isn't willing to disclose it).
The reason I say we don't need it is because I believe I have successfully mapped down the input stage of this board, and it's quite possible we won't need anything else beyond that. But for what it's worth, the output of my schematic, one side of what is probably a coupling capacitor, connects to pin 10 (IN+) of the LM3886. We may not need to go past that point. Remember the OS/CS configuration mystery?
Again, this is the part of the circuit I've mapped:
And this is the schematic:
The reason I made the schematic is to verify that I got it right using simulation. Earlier, I crudely calculated the gain the op-amp offers (by actually removing it and tapping into the output pin socket) to 4.22. Here's a transient analysis of the circuit above:
With a peak input of 100mV I get around 470mV output. Here's a frequency sweep:
It shows 13.5dbV across the frequency range, which translates to a gain of 4.73.
I think next steps are to see how gain can be manipulated using different resistor values, and most importantly if there's some magic hidden in those OS/CS jumper modes the seller mentions, related to impedance. We had a whole discussion earlier in the thread about inserting a high Zin - low Zout buffer at the input of this board and just forget it. But it's possible one of those two functions is exactly that, implementation of a current amplifier with unity gain. I 'll update my schematic and report back.
Last edited:
Good!!!
IF so, and I only see AC coupling via C1, just unsolder and lift C1 left leg (as in schematic), and inject Audio signal there, straight to the (for now) flying leg.
* You should get Audio into speaker.
* You should need about 500mV to 1V to reach clipping, the standard LM3886 sensitivity.
* You should measure about 50k impedance there, meaning a 47k series resistor should about halve speaker out level.
IF all 3 conditions are met, you have solved your problem 🙂
PS: just in case, and for safety reasons, do the test using a dim bulb limiter, a 25W to 40W bulb is recommended (a tungsten filament one, no LED, CFL, etc.) and also without speaker load, you don´t need it to measure sensitivity and impedance.
Post results.
EDIT: you can ground input signal at the other end of R5, which is a "safe" ground.
IF so, and I only see AC coupling via C1, just unsolder and lift C1 left leg (as in schematic), and inject Audio signal there, straight to the (for now) flying leg.
* You should get Audio into speaker.
* You should need about 500mV to 1V to reach clipping, the standard LM3886 sensitivity.
* You should measure about 50k impedance there, meaning a 47k series resistor should about halve speaker out level.
IF all 3 conditions are met, you have solved your problem 🙂
PS: just in case, and for safety reasons, do the test using a dim bulb limiter, a 25W to 40W bulb is recommended (a tungsten filament one, no LED, CFL, etc.) and also without speaker load, you don´t need it to measure sensitivity and impedance.
Post results.
EDIT: you can ground input signal at the other end of R5, which is a "safe" ground.
Thank you for all the pointers/info, clipping level, ground point, and especially the light bulb as a load trick. It's much quieter. 🙂
Quick question (since I'd like to dwell a little more on the NE5534 which is so much fun): I observed through simulation that if I replace all four R1-R4 with equal value resistors, I get unity gain and my input impedance is that value of resistor. I could go high with the resistor values, say 40KΩ and keep the op amp in its place. But I wonder if there are upper limits to the resistor values I can use with this network. Also, besides the unity gain, would the op amp serve some useful purpose in such a scenario, or would it be just an unnecessary stage very slightly degrading the input signal?
Last edited:
1) You´re welcome
2) I didn´t suggest the bulb as a load but in series with Mains plug, not the same by a Country Mile.
Read about dim bulb limiters and how to use them.
3) when solving an unknown problem, I try to minimize variables.
Not sure what those Op Amps do, not needing¨them, I would just avoid them. Period.
2) I didn´t suggest the bulb as a load but in series with Mains plug, not the same by a Country Mile.
Read about dim bulb limiters and how to use them.
3) when solving an unknown problem, I try to minimize variables.
Not sure what those Op Amps do, not needing¨them, I would just avoid them. Period.
😎
Another Occam's Razor: schematic post 51: replace R1 with 22k, set switch OS to on. Done.
Jan
Another Occam's Razor: schematic post 51: replace R1 with 22k, set switch OS to on. Done.
Jan
Last edited:
You beat me to it Jan, because I inadvertently revealed where the OS switch is located, earlier than I wanted in this discussion.
Yesterday I did a little RTFM of the NE5534's manual (very unusual for me) and it had this nice picture with words over it:
Initially I thought my circuit was not in a voltage follower configuration because of the existence of R4, but I was wrong (I thought R4 had to be zero - shorted by OS switch - for it to work as a voltage follower).
Obviously with R4 gone we get unity gain, which is not unwelcome. But what about other op amp characteristics? Texas Instruments mentions voltage follower as typical application so there shouldn't be many wrong aspects to it, right? However, here's what Rod Elliott (ESP) has to say about this:
The B) Non-inverting schematic shows the exact same direct connection of output to inverting input suggested. Plus it shows the necessary R1
at the non inverting input, missing from the TI schematic.
Returning to my circuit, I believe I can easily simplify by shorting R4 (OS), disconnecting R3, shorting R2, and set preferred input impedance using R1:
A layman's evaluation would be that gain is unity which is welcome, it is trivial to pick input impedance with a resistor due to the op amp's huge input impedance, output impedance is extremely low which can only be a good thing (right?), and there's current amplification at the output which sounds great (but may be or not be welcome/useful).
This all sounds good, but won't this stage suffer from the ill effects (output headroom/distortion) mentioned in Mr. Elliott's brief analysis?
(A reminder is that its output is connected to IN+ of a LM3886 amp chip.)
Last edited:
You do realize this has nothing to do with the NE5532 but everything with those things we call opamps?
BTW Your solution is no longer Occams Razor. Too much unnecessary stuff ;-)
Also, Elliot specifically mentions 'as with all opamp circuits ....'.
You also should have noticed that he mentioned a limited output current into 8 ohms which of course is not applicable here therefor irrelevant.
Not sure why you underlined that.
Jan
BTW Your solution is no longer Occams Razor. Too much unnecessary stuff ;-)
Also, Elliot specifically mentions 'as with all opamp circuits ....'.
You also should have noticed that he mentioned a limited output current into 8 ohms which of course is not applicable here therefor irrelevant.
Not sure why you underlined that.
Jan
I think I understand there will be no headroom/distortion issues related to output current when feeding an LM3886.
But why do you say too much unnecessary stuff? This circuit is even simpler than your Occam's Razor of "schematic post 51: replace R1 with 22k, set switch OS to on".
Can't resist: we have a gain block 'opamp' that has a differential input and a gain of hundreds of thousands.
That means that if the output is just a few volts, the two inputs need to be almost at the same voltage. Right?
So when we connect the inverting input to the output, and input say 1V to the non-inverting input, we know that the inverting input has to be very close to 1V too, right?
Since the output is connected to the inverting input, the output is also very close to 1V.
How much is the difference between the two inputs?
Assume a gain of 100,000, and 1V input, the difference between the two inputs is 1/100,000 = 10uV.
Easy ;-)
Jan
That means that if the output is just a few volts, the two inputs need to be almost at the same voltage. Right?
So when we connect the inverting input to the output, and input say 1V to the non-inverting input, we know that the inverting input has to be very close to 1V too, right?
Since the output is connected to the inverting input, the output is also very close to 1V.
How much is the difference between the two inputs?
Assume a gain of 100,000, and 1V input, the difference between the two inputs is 1/100,000 = 10uV.
Easy ;-)
Jan